UPDATED: Did Clovis play a deceptive game with rociletinib data in leadup to its PhIII stunner?

Kapil Dhingra

New analysis published in the Annals of Oncology raises serious questions about the truthfulness and timing involved in the way Clovis Oncology ($CLVS) corrected pivotal data on its cancer drug rociletinib.

Longtime cancer drug development expert Kapil Dhingra, who sits on a variety of biotech boards and once headed oncology drug development at Roche ($RHHBY), believes that the efficacy data for the drug was misrepresented by the company as it rushed ahead with an accelerated case to make the drug appear more effective than it actually turned out. And he believes that an upcoming FDA advisory committee group--scheduled to meet next week--will probe carefully into the company's data before recommending any approval to a rival for AstraZeneca's ($AZN) Tagrisso, which appears to have superior efficacy data.

"I feel that the efficacy data have, consistently and repeatedly, over many years, been misrepresented," Dhingra told me. "This is not simply a case of gray zones, this is black and white untrue presentation of the data. And it is not just a minor misrepresentation (such as photoshopping a western blot image etc that can get a basic scientist in trouble); the true efficacy is about half of what they represented."

With shares pumped to about $100, Clovis stunned investors last fall when it suddenly slashed the response rate on rociletinib for non-small cell lung cancer to 34% for the 625-mg dose arm, and 28% for the 500-mg dose arm, a plunge of more than 20 points. The news put the drug--positioned as a promising rival to Tagrisso (AZD9291), which is backed with peak sales projections of $3 billion--at a distinct competitive disadvantage, and the subsequent rout eviscerated the biotech's share price. Its shares closed Friday at $19.37.

Clovis says that the radical change-up was due to an unexpected, eleventh-hour erosion in the data, as the unconfirmed responses tracked in the interim data failed to pan out in the final analysis as confirmed responses. 

In the journal article, Dhingra takes issue with the way in which Clovis executives and their investigators reported the data, saying that they failed to highlight the distinction between unconfirmed and confirmed responses according to clear, well established industry trial standards, touting promising outcomes even when the data should have been sufficiently advanced to alert the company to the truth long before the big reveal.

"It is conceivable that nothing out of the ordinary happened and that we are simply looking at the natural evolution of the data," Dhingra cautions in the article. "Mature datasets often reveal the true efficacy of a new drug to be less than the efficacy observed in the initial data of on-going trials. However, a drop in response rate from 59% to 34% is quite extraordinary. So, the question is: Did the efficacy actually drop over time or was it a case of lack of full disclosure of the data, consistently and repeatedly? For decades, a complete response (CR) or partial response (PR) in a cancer patient has required, by definition, confirmation of a defined threshold of objective tumor shrinkage by a re-evaluation performed at least 4 weeks later. Any unconfirmed responses can be reported in publications with appropriate annotations but are never the primary endpoint in a trial, especially in a pivotal trial designed to seek marketing authorization."

Boulder, CO-based Clovis, though, was allowed to publish the suspect data in the prestigious New England Journal of Medicine, making a high-profile case that the biotech had to recant later. And would the NEJM have even accepted the study for publication, Dhingra asks, if the response rates were as low as they eventually turned out to be? 

Clovis CEO Patrick Mahaffy

Clovis CEO Patrick Mahaffy did not respond immediately to a query from FierceBiotech this morning. After the market closed on Monday, with the stock down 11% for the day, the company sent a statement to me saying:

"We strongly disagree with the recent editorial in the Annals of Oncology published on April 4. There will be an opportunity to discuss our research at a more robust and appropriate forum. Clovis Oncology believes in the strength of the rociletinib clinical trial program, and we look forward to upcoming discussions of our data."

Dhingra is a longtime drug developer, with stints at Eli Lilly ($LLY) as well as Roche. He currently is the principal officer at KAPital Consulting and sits on the board at Advanced Accelerator Applications, Five Prime, Exosome Diagnostics and Autolus. Dhingra held the same position at Micromet and BioVex, both bought out by Amgen in deals worth up to $2.6 billion. He is also an associate editor at the Annals of Oncology, though he wrote the piece on Clovis as an individual expert observer.

The author--who emphasized that he has no business relationship with either AstraZeneca or Clovis--adds that he followed the tale of the T790 EGFR inhibitors as a "great example of targeted and personalized medicine, sort of the can't-miss drugs. Of course, the reality has played out quite differently."

And he's deeply rankled by what happened at Clovis.

Sticking with what is solely a matter of public record, Dhingra says Clovis's timeline on reporting drug responses may not stand up under expert scrutiny.

"None of the (company's) three press releases disclosed as to what proportion of responses, if any, were confirmed responses according to the usual and customary definition of response," he writes. "It is worth emphasizing that the ASCO presentation was only about two months before the completion of the company's NDA filing. Importantly,  majority of the patients in the TIGER-X trial had been enrolled in 2014 or earlier (Clovis Oncology ASCO 2015 investors' event). This raises a very distinct possibility that by ASCO 2015, the data from the TIGER-X trial were relatively mature with respect to confirmation of responses."

Dhingra critiqued the way Clovis sought to turn an unexpected side effect, hyperglycemia, as being potentially beneficial based on a preclinical rationale, when clinical data from more advanced drugs had cast serious doubts on the validity of the hypothesis. He also questioned the switch in the company's position on the optimal dose of the drug over a relatively short time period. He states: "after treating over 600 patients, the optimal dose of rociletinib is still not clear."

"Drugs that are touted as being highly selective for a particular target, often hit other targets or lead to off-target toxicity," he notes. "In this case, hyperglycemia was a meaningful clinical and pharmacological effect and in the end contributed to the overall safety and efficacy profile. Limited information has been presented by the sponsor at public meetings that attributes it to a metabolite of the parent drug, which inhibits type I insulin-like growth factor receptor and, to a lesser extent, insulin receptor kinases in biochemical and cellular studies. A fuller presentation and discussion of such data is warranted when investigators present the study data in important scientific congresses. Illogical rationalizations to turn negative data to positive attributes should be avoided."

The lower final response rate may make the drug uncompetitive against AstraZeneca's therapy, but that wouldn't necessarily prevent the FDA or its advisory group from endorsing the drug for marketing. For several years now the FDA's oncology advisory group has been very receptive to any therapies that can add another arrow to an oncologist's quiver in the fight to delay cancer. And while lower response rates may be less appealing, the data suggest the drug is having an impact, which would normally be enough for an approval.

But if the outside experts end up questioning how Clovis handled the data, and its overall trustworthiness in what was finally offered up, a cloud of uncertainty could threaten an FDA rejection, particularly if it has already riled agency cancer czar Richard Pazdur. FierceBiotech has reported on a number of cases where Pazdur forcefully stepped in to raise objections to an application during a review, including a panel appearance for Aveo, which was later forced to settle SEC charges that it deceived investors on the FDA's trial demands.

During an interview with me, Dhingra underscored his concerns that Clovis's actions could raise questions--and regulatory fears--about all accelerated development efforts in the cancer field. And that could blight other biotech's chances to pursue the kind of fast-paced development schedules that have become the norm.

"In this era of rapidly evolving science and breakthrough medicines, we are all rushing full speed ahead to get important medicines to patients quickly," Dhingra sums up. "The story of rociletinib should serve as a reminder that sometimes 'haste makes waste'. This author has forcefully argued for acceleration of targeted oncology drugs based on relatively small datasets from early clinical trials. Such a thoughtful acceleration of drug development is possible in a variety of ways but it requires a vigilant attitude toward emerging data and quick course corrections in the event new data are not consistent with the initial observations. These lessons will become increasing important especially as we move deeper and further into the emerging world of novel therapies, especially immuno-oncology and cellular therapies."

- here's the report